newegg-pc-builder

Compare original and translation side by side

🇺🇸

Original

English

🇨🇳

Translation

Chinese

Newegg PC Builder MCP Skill

Connects Claude to the Newegg PC Builder MCP service. This skill is fully dynamic: it discovers available tools at runtime and lets the LLM decide which tool to call and how to fill its parameters. No tool names or parameter names are hard-coded, so the skill continues to work even after the MCP server updates its API.

Required header on every request: all calls to

apis.newegg.com/ex-mcp/...

must carry

x-skill: newegg-pc-builder

in addition to

Content-Type

. It identifies the calling skill to the endpoint — include it even when you assemble a request by hand rather than copying an example below.

MCP Endpoint:

https://apis.newegg.com/ex-mcp/endpoint/pcbuilder

Script:

scripts/mcp_client.py

将Claude连接至Newegg PC Builder MCP服务。此Skill是完全动态的：它会在运行时发现可用工具，让LLM决定调用哪个工具以及如何填充参数。工具名称或参数名称均未硬编码，因此即使MCP服务器更新其API，该Skill仍能正常工作。

每个请求的必填请求头：所有对

apis.newegg.com/ex-mcp/...

的调用，除

Content-Type

外，还必须携带

x-skill: newegg-pc-builder

。它用于向端点标识调用方Skill——即使你手动组装请求而非复制下方示例，也需包含此请求头。

MCP端点：

https://apis.newegg.com/ex-mcp/endpoint/pcbuilder

脚本：

scripts/mcp_client.py

Core Workflow (always follow this order)

核心工作流程（请始终遵循此顺序）

Step 1 — Discover tools

步骤1 — 发现工具

Always start by listing available tools. Never assume tool names or parameters from previous runs or documentation.

bash

python scripts/mcp_client.py list_tools

The output contains, for each tool:

```
name
```
— identifier to use when calling
```
description
```
— what it does (may be empty; infer from name + schema)
```
inputSchema.properties
```
— available parameters with types and descriptions
```
inputSchema.required
```
— mandatory parameters

始终先列出可用工具。切勿假设工具名称或参数与之前运行或文档中的一致。

bash

python scripts/mcp_client.py list_tools

输出包含每个工具的以下信息：

```
name
```
— 调用时使用的标识符
```
description
```
— 工具功能（可能为空；可从名称+架构推断）
```
inputSchema.properties
```
— 可用参数及其类型和描述
```
inputSchema.required
```
— 必填参数

Step 2 — Select tool and map parameters

步骤2 — 选择工具并映射参数

Read the

list_tools

output and decide:

Which tool best matches the user's intent?
- Match on description first; fall back to inferring from name + param names
- If multiple tools apply, prefer the most specific one
- If still ambiguous, pick the first one and note the assumption
How does user intent map to parameters?
- Only use parameters present in
```
inputSchema.properties
```
- Free-text params (e.g.
```
question
```
  ,
```
query
```
  ,
```
text
```
  ): pass the user's request as a natural-language string describing their need
- Typed/enum params: map user intent to the closest valid value
- Leave optional params unset unless you have a clear value
- Never invent parameters not present in the schema

阅读

list_tools

的输出并决定：

哪个工具最符合用户意图？
- 优先匹配描述；若描述为空，可从工具名称+参数名称推断
- 若多个工具适用，选择最具体的一个
- 若仍存在歧义，选择第一个工具并注明此假设
用户意图如何映射到参数？
- 仅使用
```
inputSchema.properties
```
  中存在的参数
- 自由文本参数（如
```
question
```
  、
```
query
```
  、
```
text
```
  ）：将用户请求作为描述其需求的自然语言字符串传递
- 类型/枚举参数：将用户意图映射到最接近的有效值
- 可选参数若无明确值则留空
- 切勿创建架构中不存在的参数

Step 3 — Call the tool

步骤3 — 调用工具

bash

python scripts/mcp_client.py call <tool_name> '<json_arguments>'

Tool name and arguments are determined at runtime from Step 2. Example:

bash

python scripts/mcp_client.py call v2allin '{"question": "gaming PC RTX 5090 9800X3D best price"}'

Windows PowerShell (important)
PowerShell parses outer double quotes before Python runs. Using

\"...\"

inside

"..."

often breaks the JSON string (you may see errors like

Got: {\

or “invalid JSON”). Prefer one of:

Single-quote the whole JSON (no backslash escapes needed):

powershell

python scripts/mcp_client.py call v2allin '{"question": "gaming PC RTX 3070"}'

JSON in a file (most reliable for long or nested payloads):

powershell

Set-Content -Path args.json -Encoding utf8 '{"question": "gaming PC RTX 3070"}'
python scripts/mcp_client.py call v2allin @args.json

Stdin (pipe or redirect):

powershell

'{"question": "gaming PC RTX 3070"}' | python scripts/mcp_client.py call v2allin -

The script supports

@path\to\file.json

and

for stdin so shells never need to escape inner double quotes.

Where to change things

Skill + script (recommended): keep examples and
```
mcp_client.py
```
in sync — document PowerShell rules here, and use
```
@file
```
/
```
-
```
in the client to avoid quoting bugs.
Repo-only docs do not fix agents that load this skill from
```
~/.agents
```
; updating the skill is the right place for portable behavior.

bash

python scripts/mcp_client.py call <tool_name> '<json_arguments>'

工具名称和参数由步骤2在运行时确定。示例：

bash

python scripts/mcp_client.py call v2allin '{"question": "gaming PC RTX 5090 9800X3D best price"}'

Windows PowerShell（重要提示）
PowerShell会在Python运行前解析外层双引号。在

"..."

内使用

\"...\"

通常会破坏JSON字符串（你可能会看到

Got: {\

或“无效JSON”等错误）。推荐使用以下方式之一：

用单引号包裹整个JSON（无需反斜杠转义）：

powershell

python scripts/mcp_client.py call v2allin '{"question": "gaming PC RTX 3070"}'

将JSON存入文件（对于长或嵌套的负载最可靠）：

powershell

Set-Content -Path args.json -Encoding utf8 '{"question": "gaming PC RTX 3070"}'
python scripts/mcp_client.py call v2allin @args.json

通过标准输入（Stdin）传递（管道或重定向）：

powershell

'{"question": "gaming PC RTX 3070"}' | python scripts/mcp_client.py call v2allin -

脚本支持

@path\to\file.json

和

（表示标准输入），因此Shell无需转义内部双引号。

修改建议

Skill + 脚本（推荐）：保持示例与
```
mcp_client.py
```
同步——在此处记录PowerShell规则，并在客户端中使用
```
@file
```
/
```
-
```
以避免引号错误。
仅仓库文档无法修复从
```
~/.agents
```
加载此Skill的代理；更新Skill才是实现可移植行为的正确方式。

Step 4 — Interpret the response

步骤4 — 解析响应

Parse the JSON output. Common response shapes:

json

{ "result": { "summary": "...", "popular": [...], "valued": [...] } }

```
summary
```
non-null → use as the primary answer text
```
popular
```
/
```
valued
```
arrays present → list the builds
All result fields null → service returned no data; tell the user and offer to answer from general knowledge instead
```
isError: true
```
→ report the error and suggest retry

解析JSON输出。常见响应格式：

json

{ "result": { "summary": "...", "popular": [...], "valued": [...] } }

```
summary
```
非空 → 将其作为主要回答文本
存在
```
popular
```
/
```
valued
```
数组 → 列出装机配置
所有结果字段均为空 → 服务未返回数据；告知用户并提议改用通用知识回答
```
isError: true
```
→ 报告错误并建议重试

Step 5 — Present results

步骤5 — 展示结果

Build list: name, total price, key components, brief description
Compatibility check: clear yes/no with reasoning
Component details: specs, price, compatibility notes
No results from API: explain clearly, then offer a manual recommendation

装机配置列表：名称、总价、核心组件、简要描述
兼容性检查：明确的是/否结论及理由
组件详情：规格、价格、兼容性说明
API无结果：清晰说明情况，然后提供手动推荐

Edge case handling

边缘情况处理

Situation	Action
`list_tools` returns 0 tools	Report service unavailable; answer from training knowledge
Tool description is empty	Infer purpose from name + parameter names
Schema has no `required`	Treat all params as optional; pass what you have
All response fields null	No data for this query; say so and fall back to general knowledge
HTTP error on `list_tools`	Report error; do not proceed to call step
HTTP error on `call`	Retry once; if still failing, fall back to general knowledge
Multiple tools match	Pick most specific; briefly note the choice

场景	操作
`list_tools` 返回0个工具	报告服务不可用；使用训练知识回答
工具描述为空	从工具名称+参数名称推断用途
架构无 `required` 字段	将所有参数视为可选；传递已有参数
所有响应字段为空	无此查询的数据；告知用户并 fallback 到通用知识
`list_tools` 请求出现HTTP错误	报告错误；不进入调用步骤
`call` 请求出现HTTP错误	重试一次；若仍失败，fallback到通用知识
多个工具匹配	选择最具体的工具；简要说明此选择

Script reference

脚本参考

scripts/mcp_client.py

uses Python standard library only (3.6+, no pip needed). Handles both

application/json

and

text/event-stream

responses automatically.

python scripts/mcp_client.py list_tools
    → prints all tools with full parameter schemas

python scripts/mcp_client.py call <tool_name> '<json_args>'
    → calls the tool and prints the JSON response

python scripts/mcp_client.py call <tool_name> @args.json
    → reads JSON arguments from a UTF-8 file (good on Windows)

echo '<json>' | python scripts/mcp_client.py call <tool_name> -
    → reads JSON from stdin

Errors go to stderr with a non-zero exit code. Invalid JSON prints a short hint for PowerShell users.

scripts/mcp_client.py

仅使用Python标准库（3.6+，无需pip安装）。可自动处理

application/json

和

text/event-stream

响应。

python scripts/mcp_client.py list_tools
    → 打印所有工具及完整参数架构

python scripts/mcp_client.py call <tool_name> '<json_args>'
    → 调用工具并打印JSON响应

python scripts/mcp_client.py call <tool_name> @args.json
    → 从UTF-8文件读取JSON参数（适用于Windows）

echo '<json>' | python scripts/mcp_client.py call <tool_name> -
    → 从标准输入读取JSON

错误会输出到stderr并返回非零退出码。无效JSON会向PowerShell用户打印简短提示。